Semester
Fall
Date of Graduation
2024
Document Type
Thesis
Degree Type
MS
College
Statler College of Engineering and Mineral Resources
Department
Lane Department of Computer Science and Electrical Engineering
Committee Chair
Jeremy Dawson
Committee Member
Gianfranco Doretto
Committee Member
Prashnna Gyawali
Abstract
This work focuses on a deblurring network designed specifically for the task of deblurring contactless fingerprints, a.k.a fingerphotos. This network takes the general idea of style transfer, and applies it to the realm of deblurring. The standard use case for style transfer networks is artistic style transfer, which is used for recreational purposes. For this work, though, style transfer is used in the context of deblurring. Since style transfer can transfer artistic styles from one image to another, who's to say it can’t transfer clarity and sharpness as well? This is the focus of this work; taking a blurry fingerphoto, and deblurring it using the sharpness and clarity found in a sharp fingerphoto. To accomplish this feat, the Discrete Wavelet Transform is used in order to split images into their respective directionally-specific sub-bands. Then, each blurry and sharp sub-band is combined, resulting in a deblurred set of sub-bands. Finally, these sub-bands are sent to a neural decoder for reconstruction and refinement. The result is a deblurred version of the original blurry fingerphoto, with minimal loss of fine details and minutiae.
Recommended Citation
Keaton, David Connard, "Fingerphoto Deblurring using Wavelet Style Transfer" (2024). Graduate Theses, Dissertations, and Problem Reports. 12690.
https://researchrepository.wvu.edu/etd/12690
